This is a revision of application 1 R01 AA015385-01 that focuses on the evaluation of a promising pharmacological intervention for patients with alcoholism complicated by comorbid bipolar disorder, an area of major unmet treatment needs. We propose a double-blind, placebo-controlled, randomized, parallel group trial to test the efficacy of the combination of the opioid antagonist, naltrexone, and the antikindling mood stabilizing agent valproate, versus valproate alone, in the treatment of patients with comorbid alcoholism and bipolar disorder. With nearly two million affected individuals in the U.S., comorbid alcoholism and bipolar disorder represents a significant public health challenge. This comorbidity is associated with severe disability, morbidity, and heightened risk for suicide. Surprisingly, little research and limited evidence-based treatment options exist for this high-risk population. Our recently published randomized controlled trial of valproate efficacy in bipolar alcoholics remains the only such study completed to date (Arch Gen Psychiatry 2005; 62:37-45). The results of that study suggested an advantage of valproate over placebo in reducing heavy alcohol use. However, a significant proportion of valproate treated subjects continued to consume alcohol at abusive levels. There are compelling theoretical, and accruing clinical evidence suggesting that combined valproate + naltrexone may have synergistic effects on decreasing alcohol use. Valproate, may decrease alcohol use by stabilizing pathological mood states, and by dampening the negative reinforcing effects of acute and protracted alcohol withdrawal. Conversely, naltrexone would decrease the positive reinforcing effects of alcohol by decreasing the desire to drink alcohol. The results of our open-label, randomized, pilot study suggests that valproate + naltrexone robustly enhances abstinence from alcohol, decreases craving, and improves mood and functioning. All are particularly desirable outcomes in patients suffering from severe psychopathology. These results provide compelling evidence for testing this approach in a randomized, controlled trial. We propose the following aims: 1) Examine the efficacy of naltrexone plus valproate compared to valproate and placebo in the treatment of patients with DSM-IV alcohol dependence and comorbid bipolar I disorder; 2) Assess the effects of primary vs. secondary alcoholism, bipolar subtype (depressed vs. manic/mixed subtype), and the presence of another substance use disorder (SUD) as moderators of alcohol use outcome; 3) Assess the effects of medication compliance, persistence of mood symptoms or SUD, and social support as mediators of alcohol use outcome. One hundred and four acutely ill and actively drinking adult subjects will be randomized and prospectively followed during a 3-month double-blind study, and a 3-month follow-up phase. All subjects will receive individual counseling designed to enhance treatment adherence.